The planetary transmission system has been widely used in industry because of its various advantages. And the study on degradation feature extraction method of planetary gearbox is of major significance for mechanical system prognostics and health management (PHM). In this paper, the alpha stable distribution characteristics of planetary gearbox vibration signals in performance degradation process are verified. By observing the change of alpha stable distribution for planetary gearbox degradation experiment data, a new degradation feature extraction method based on alpha stable distribution is proposed, which is called the height of probability distribution (HPD). Through comparative analysis, it is determined that HPD has better linearity and less fluctuation compared with conventional degradation features in planetary gearbox accelerated degradation stage. Moreover, in the accelerated degradation stage, the degradation trend prediction result based on HPD is closer to the actual data than conventional degradation features no matter using Wiener-based or LSSVM-based prediction method. These conclusions indicate that the newly proposed HPD works well and gives accurate estimates for condition monitoring and degradation trend prediction of planetary gearbox.

行星传动系统由于其各种优点而已在工业中广泛使用。行星齿轮箱退化特征提取方法的研究对机械系统的故障预测和健康管理具有重要意义。本文验证了行星齿轮箱振动信号在性能下降过程中的α稳定分布特性。通过观察行星齿轮箱退化实验数据的α稳定分布变化，提出了一种基于α稳定分布的退化特征提取新方法，即概率分布高度（HPD）。经过比较分析 与行星齿轮箱加速退化阶段的常规退化特征相比，HPD具有更好的线性度和较小的波动。此外，在加速退化阶段，无论使用基于Wiener的预测方法还是基于LSSVM的预测方法，基于HPD的退化趋势预测结果都比常规退化特征更接近实际数据。这些结论表明，新提出的HPD可以很好地工作，并且可以为行星齿轮箱的状态监测和退化趋势预测提供准确的估计。